Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754245Ab3COMMc (ORCPT ); Fri, 15 Mar 2013 08:12:32 -0400 Received: from zeniv.linux.org.uk ([195.92.253.2]:59680 "EHLO ZenIV.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754075Ab3COMMb (ORCPT ); Fri, 15 Mar 2013 08:12:31 -0400 Date: Fri, 15 Mar 2013 12:12:20 +0000 From: Al Viro To: James Bottomley Cc: "J. R. Okajima" , Miklos Szeredi , Andrew Morton , torvalds@linux-foundation.org, linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, hch@infradead.org, apw@canonical.com, nbd@openwrt.org, neilb@suse.de, jordipujolp@gmail.com, ezk@fsl.cs.sunysb.edu, dhowells@redhat.com, sedat.dilek@googlemail.com, mszeredi@suse.cz Subject: Re: [PATCH 0/9] overlay filesystem: request for inclusion (v17) Message-ID: <20130315121220.GY21522@ZenIV.linux.org.uk> References: <1363184193-1796-1-git-send-email-miklos@szeredi.hu> <20130313160854.54ac0491044371b4db214698@linux-foundation.org> <20130315012541.GU21522@ZenIV.linux.org.uk> <19058.1363320936@jrobl> <20130315044411.GW21522@ZenIV.linux.org.uk> <20079.1363324154@jrobl> <20130315051322.GX21522@ZenIV.linux.org.uk> <1363335318.2459.4.camel@dabdike> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1363335318.2459.4.camel@dabdike> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1393 Lines: 23 On Fri, Mar 15, 2013 at 08:15:18AM +0000, James Bottomley wrote: > It's the fact that a directory entry based whiteout limits the amount of > change to the VFS, but has to be supported by underlying filesystems. > The generic_dirent_fallthrough() mechanism is a nice way of hiding it, > but there are still quite a few fs specific mods in the union mount tree > because of this. Having to modify filesystems to me indicates the > mechanism is a bit fragile. If we could do whiteouts purely in the VFS, > so it would work for any filesystem (without needing filesystem > modifications) that would seem to be a more robust approach. I'm not > saying we can definitely do this in an elegant way ... I'm just saying > that if someone comes up with it, it's obviously preferable. The trouble with such mechanisms is that they tend to end up depending on fairly non-trivial properties of underlying fs. Try aufs one on btrfs, see how soon you spot the problem. It's nice when a method turns out to be really redundant and implementable in uniform way via other methods present; see fh_to_dentry history for example of situation where it hadn't... -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/